Various methods have been proposed as technology for predicting the profile of a metal strip, such as a sheet or a plate, after rolling.
For example, Japanese Patent Application Laid-Open (JP-A) No. 2008-112288 describes technology that improves the prediction precision for an extrapolation region for which actual data does not exist, and also corrects errors in a rolling model. Specifically, a database of actual results, in which manufacturing conditions of previously manufactured products are stored associated with manufacture outcome information, is employed to compute a degree of similarity between respective samples in the database of actual results and request points (prediction target points), and to generate a prediction formula for the vicinity of the request points using weighted regression weighted by the degree of similarity. The prediction precision for the extrapolation region is improved by the prediction formula.
JP-A No. 2005-153011 describes technology that predicts the profile of a metal strip by splitting elongation strain (stress) that is distributed in a strip width direction of a metal strip during rolling into elongation strain that is geometrically transformed into a wave profile during buckling, and elongation strain still present in the metal strip after buckling.
Moreover, JP-A No. 2012-218010 describes technology that predicts the profile of a metal strip by measuring characteristic amounts of the profile of the metal strip at exit from a rolling mill, and also finding elongation strain present in the metal strip during measurement, then superimposing the elongation strain on the profile characteristic amounts, and measuring this as true profile characteristic amounts applied by the rolling mill. Note that positions in a strip passing direction of the strip and a width direction of the strip, and height direction displacement, are measured on exit from the rolling mill as geometric values. Moreover, profile, steepness, and elongation strain difference are found as the profile characteristic amounts.